Trackers input type unclear

[YiannisParask]

In the following block of code you ask for a numpy array of detections in the format [[x1,y1,x2,y2,score],[x1,y1,x2,y2,score],...] and later on you try to convert it to numpy again. Wouldn't that raise an error?

xyxys = dets[:, 0:4]
        scores = dets[:, 4]
        clss = dets[:, 5]
        
        classes = clss.numpy()
        xyxys = xyxys.numpy()
        scores = scores.numpy()

[mikel-brostrom]

What should be passed to all trackers is torch arrays. I need to find time to fix the docstrings.

[YiannisParask]

Thanks for answering! My knowledge of torch is very limited and i've been trying to figure out a solutions for days.

[mikel-brostrom]

Just an update. All inputs are now numpy.ndarray. Check out: https://github.com/mikel-brostrom/yolo_tracking#custom-object-detection-model-example

[YiannisParask]

Thanks for the update 😃